Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 
World Community Grid Forums
  Category: Active Research
  Forum: OpenPandemics - COVID-19 Project
  Thread: OpenPandemics - GPU Stress Test

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 781
Posts: 781   Pages: 79   [ Previous Page | 24 25 26 27 28 29 30 31 32 33 | Next Page ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 585115 times and has 780 replies Next Thread
Richard Haselgrove
Senior Cruncher
United Kingdom
Joined: Feb 19, 2021
Post Count: 360
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
Trying to load up a second high-throughput Linux machine, but got these messages from the server - I haven't seen the second from other projects.
27/04/2021 16:38:05 | World Community Grid | No tasks are available for OpenPandemics - COVID 19
27/04/2021 16:38:05 | World Community Grid | Tasks are committed to other platforms

----------------------------------------
[Edit 1 times, last edit by Richard Haselgrove at Apr 27, 2021 3:43:02 PM]
[Apr 27, 2021 3:42:25 PM]   Link   Report threatening or abusive post: please login first  Go to top 
m0320174
Cruncher
Joined: Feb 13, 2021
Post Count: 11
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
Thanks! So I am not alone. So if I understand correctly, I would need much more CPU power for a task to keep the GPU busy all the time. (Or maybe it is even impossible?).

I'm not sure if it's possible to continuously max out your GPU. I see 2 reasons why it can be a challenge:
1. the CPU needs to "prep" the workunit before the GPU can actually start to work on it. During this time your GPU is completely idle (as mentioned above).
2. if you have a powerful GPU then 1 workunit will not max out your GPU

Some people run multiple GPU workunits in parallel even if they only have 1 GPU. I have not experimented with this.
[Apr 27, 2021 3:46:58 PM]   Link   Report threatening or abusive post: please login first  Go to top 
goben_2003
Advanced Cruncher
Joined: Jun 16, 2006
Post Count: 145
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
Does the continous disk writing i see people complaining about happen for temporary files? I'm not experiencing it, but i have my temporary folder (tmp) mounted in ramdisk through tmpfs, and i only see on the SSD the occasional "dumps" of data, every few minutes. Can this be it?


I wouldn't worry about it. the fears of SSD writes are largely FUD and blown out of proportion. most modern SSDs can handle PETABYTES of writes before failure is a concern and they have more advanced wear leveling than earlier SSDs. that's continuous writing for 10+ years in most cases. and real world use will be far below that.

Sorry Ian-n-Steve, I am going to have to disagree. It is not FUD or blown out of proportion. I think that there is a very wide distribution of total writes ratings across SSDs. Using the samsung 860 2tb and crucial mx500 2tb as examples:
Total bytes written ratings:
samsung: 2400 TB(2.4PB)
crucial: 700 TB
max MB/s to reach 10 years:
samsung: ~7.61
crucial: ~2.22

Some people are seeing higher writes than either of those SSDs are rated for to last 10 years. I do not currently have server grade SSDs like nanoprobe has. There is a reason I run boinc off of a ramdisk these days - and it is not FUD. tongue
----------------------------------------
[Apr 27, 2021 4:03:53 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Grumpy Swede
Master Cruncher
Svíþjóð
Joined: Apr 10, 2020
Post Count: 2186
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
OK, the biggies are starting to validate for me now.
[Apr 27, 2021 4:25:17 PM]   Link   Report threatening or abusive post: please login first  Go to top 
William Albert
Cruncher
Joined: Apr 5, 2020
Post Count: 39
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
A few things I've noticed during this test on my machines:

  • Machines in my fleet with Nvidia GPUs are being supplied a steady stream of WUs.
  • WU availability for Intel GPUs is inconsistent. I have four machines with idle Intel GPUs that are available but not getting work, even though two other Intel GPU machines are getting a steady stream of WUs.
  • My one machine with an AMD GPU seems to be getting a steady stream of work for that GPU.
  • I have two machines with Nvidia GT 720 GPUs that have pulled more Nvidia WUs than they are likely to be able to complete by the reporting deadline. One of them intermittently fails WUs due to them taking too long.
  • Machines with multiple GPU brands (e.g., a discrete Nvidia GPU and an integrated Intel GPU) are crunching WUs for both without any problems, at least on Linux.
  • The two machines with slow Nvidia GPUs with WUs that intermittently time out also have Intel GPUs that are crunching fine, but I don't have an obvious way of setting NNT for just the Nvidia GPU. There's probably a way to do it with some profile trickery.

Overall, a fun test so far. biggrin
----------------------------------------
[Edit 1 times, last edit by William Albert at Apr 27, 2021 4:34:58 PM]
[Apr 27, 2021 4:34:02 PM]   Link   Report threatening or abusive post: please login first  Go to top 
JohnDK
Advanced Cruncher
Denmark
Joined: Feb 17, 2010
Post Count: 77
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
Don't know how often it happens, but I've noticed several times those 5 digit WUs reach around 50% progress and then jump back to 0% and continues. Is that normal behaviour?
----------------------------------------
Intel i7-6850K / 16GB / RTX 3090 / 2x RTX 3080 Ti / RTX 3070 Ti
AMD Ryzen 9 5950X / 32GB / RTX 2080 Ti
[Apr 27, 2021 4:48:55 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Andyman
Cruncher
Joined: Apr 9, 2021
Post Count: 17
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
Don't know how often it happens, but I've noticed several times those 5 digit WUs reach around 50% progress and then jump back to 0% and continues. Is that normal behaviour?

Yeah the % seems wonky atm. Goes up and down but the timer stays the same?
It finishes as normal it seems.
[Apr 27, 2021 4:53:36 PM]   Link   Report threatening or abusive post: please login first  Go to top 
goben_2003
Advanced Cruncher
Joined: Jun 16, 2006
Post Count: 145
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: OpenPandemics - GPU Stress Test
  • Machines in my fleet with Nvidia GPUs are being supplied a steady stream of WUs.

    • I wish that were the case still for me! I finished the last Nvidia GPU WUs an hour ago, now I only have and get Intel WUs. Not complaining though, I was crunching several at a time before.
    ----------------------------------------
    [Apr 27, 2021 4:55:26 PM]   Link   Report threatening or abusive post: please login first  Go to top 
    gordonbb
    Cruncher
    Canada
    Joined: May 14, 2019
    Post Count: 19
    Status: Offline
    Project Badges:
    Reply to this Post  Reply with Quote 
    Re: OpenPandemics - GPU Stress Test
    Sorry Ian-n-Steve, I am going to have to disagree. It is not FUD or blown out of proportion. I think that there is a very wide distribution of total writes ratings across SSDs. Using the samsung 860 2tb and crucial mx500 2tb as examples:
    Total bytes written ratings:
    samsung: 2400 TB(2.4PB)
    crucial: 700 TB
    max MB/s to reach 10 years:
    samsung: ~7.61
    crucial: ~2.22

    Some people are seeing higher writes than either of those SSDs are rated for to last 10 years. I do not currently have server grade SSDs like nanoprobe has. There is a reason I run boinc off of a ramdisk these days - and it is not FUD.
    The other option with consumer grade SSDs is to use the software from the manufacturer to "over-provision" the space reserved on the SSD for wear-leveling. You start out with a smaller effective disk size but, in theory, it will last longer.

    Here is a good article from Samsung on this subject
    ----------------------------------------

    AMD - 2600x, 2 x 2700, 2700x, 3900x, 3950x, 2 x 5900x, 5950x
    Intel - E3-1231v3, 9900K
    NVidia - GTX 1060 6GB, 1660ti, 1070ti; RTX 2060, 2060s, 2070a, 5 x 2070s
    [Apr 27, 2021 4:58:03 PM]   Link   Report threatening or abusive post: please login first  Go to top 
    spRocket
    Senior Cruncher
    Joined: Mar 25, 2020
    Post Count: 274
    Status: Offline
    Project Badges:
    Reply to this Post  Reply with Quote 
    Re: OpenPandemics - GPU Stress Test
    Don't know how often it happens, but I've noticed several times those 5 digit WUs reach around 50% progress and then jump back to 0% and continues. Is that normal behaviour?

    Yeah the % seems wonky atm. Goes up and down but the timer stays the same?
    It finishes as normal it seems.


    The jump back is where the CPU hands the work off to the GPU and the real crunching begins. I'm using the boinctui-extended text user interface to manage multiple systems, and it also shows the switchover by changing its memory usage value from something reasonable to 40+ GB, which is more memory than I have in the machine and GPU combined. That might be some sort of bug in the TUI, but it does indicate when the GPU is doing the actual work.
    [Apr 27, 2021 5:04:30 PM]   Link   Report threatening or abusive post: please login first  Go to top 
    Posts: 781   Pages: 79   [ Previous Page | 24 25 26 27 28 29 30 31 32 33 | Next Page ]
    		[ Jump to Last Post ]
    		Post new Thread